SUMMARY Acoustic stimuli within the sonic range are effective triggers of C-type escape behaviours in fish. We have previously shown that fish have an acute sensitivity to infrasound also, with acceleration thresholds in the range of 10 –5 m s –2 . In addition, infrasound at high intensities around 10 –2 m s –2 elicits strong and sustained avoidance responses in several fish species. In the present study, the possible triggering of C-escapes by infrasonic single-cycle vibrations was examined in juvenile roach Rutilus rutilus . The fish were accelerated in a controlled and quantifiable manner using a swing system. The applied stimuli simulated essential components of the accelerations that a small fish would encounter in the hydrodynamic flow field produced by a predatory fish. Typical C- and S-type escape responses were induced by accelerations within the infrasonic range with a threshold of 0.023 m s –2 for an initial acceleration at 6.7 Hz. Response trajectories were on average in the same direction as the initial acceleration. Unexpectedly, startle behaviours mainly occurred in the trailing half of the test chamber, in which the fish were subjected to linear acceleration in combination with compression, i.e. the expected stimuli produced by an approaching predator. Very few responses were observed in the leading half of the test chamber, where the fish were subjected to acceleration and rarefaction, i.e. the stimuli expected from a suction type of predator. We conclude that particle acceleration is essential for the directionality of the startle response to infrasound, and that the response is triggered by the synergistic effects of acceleration and compression.

1. Recordings were made from 93 single units in the acoustical lobes and the torus semicircularis of the cod during sound stimulation and anodal blocking of the posterior saccular nerves. 2. Most medullary sound responses were phase locked to the stimulus to some degree. The phase locking was less pronounced in the torus semicircularis , and sound stimulation sometimes caused clear inhibition of activity in this area. A large fraction of the units in both recording loci was insensitive to our sound stimuli, which acted mainly via the swimbladder. 3. Peripheral blocking caused decreased activity and inhibition of sound responses in the acoustic lobes, indicating excitatory ascending input to this region. Binaural interaction was found in 8 of 29 medullary units tested during both ipsi- and contralateral block. 4. Single-sided blocking experiments revealed both inhibitory and excitatory input to the torus semicircularis region. Binaural interaction was found in 3 of the 5 units tested during both ipsi- and contralateral block in this area.

1. The nervous activity of single auditory neurones in goldfish brain have been measured. 2. Four types of acoustic stimuli were used, (1) pure tones, (2) noise of one-third octave band width, (3) noise of one-octave band width with centre frequency equal to the pure tone, and (4) white noise. 3. Except for white noise, these stimuli produced the same response to equal sound pressures. The white noise response was less, presumably because the frequency range covered by a single neurone is far narrower than the range of white noise. 4. The conclusion has been reached that for low-frequency acoustic signals, the acoustic power over a frequency band of one to two octaves is integrated by the nervous system. 5. The masking effect of background noise on the acoustic threshold of single units to pure tones is strongest when the noise band has the same centre frequency as the test tone. In this case the tone threshold increases linearly with the background noise level. 6. When the noise band was centred at a different frequency from the tone, the masking effect decreased at a rate of 20-22 dB/octave for the first one-third octave for a tone frequency of 250 Hz. For a tone of 500 Hz the masking effect of lower frequencies was stronger and was reduced by only some 9 dB/octave for the first one-third octave.

1. Saccular microphonic potentials in cod ( Gadus morhua ) were recorded by means of implanted electrodes during sound stimulation with different swimbladder volumes. To obtain acceptable acoustic conditions, the experiments were conducted at 6 m depth in the sea. 2. Swimbladder volume had no effect on the microphonic potentials at 100 Hz, whereas its effect was marked at higher frequencies. The sound pressure necessary to evoke microphonic potentials just above the electric background noise was about 20 dB higher for the empty bladder, compared to the full bladder at 300 Hz. 3. The microphonic potentials are ultimately linked to excitation of eighth-nerve fibres, and it is concluded that the existence of gas in the cod swimbladder has a positive effect on hearing sensitivity for all frequencies from a lower transient frequency to the upper frequency limit of hearing. Gas in the swimbladder will furthermore extend the audible frequency range. 4. The upper frequency limit for the saccular microphonic potentials is in good agreement with the upper audible frequency limit determined from behavioural audiograms for cod, thus indicating that this frequency cut-off may be due to inadequacy of the peripheral auditory apparatus.

1. Sound reception in the Brazilian cicada Fidicina rana has been studied by recording the electrical activity in the auditory nerve in response to various sound stimuli. 2. The highest sensitivity to pure tones is in the frequency range of 6-9 kHz. A considerable and rapid adaptation occurs. 3. The response to trains of clicks consists of a series of compound action potentials. At repetition rates above 100/sec. this response also shows adaptation. 4. The response to the natural continuous song of the cicada is not very different from the response to trains of artificially produced clicks. In the two other types of natural song the adaptation of the auditory response is circumvented. Thus the response to the ‘zeep’ call (during which the intensity of consecutive clicks increases) consists of a series of compound action potentials of about the same amplitude, and the response to the distress call consists of a 8hort burst of action potentials for each burst of clicks. It is suggested that these songs therefore constitute very effective methods of intraspecific signalling.